编译原理实验（一）——词法分析

不念不忘少年蓝@ 2023-05-30 09:15 23阅读 0赞

### 编译原理实验（一）——词法分析 ###

*   *   *  实验要求
         *  参考程序
         *  运行截图

### 实验要求 ###

输入为字符串（或待进行词法分析的源程序），输出为单词串，即由（单词，类别）所组成的二元组序列；

### 参考程序 ###

#include <string>
    #include <iostream>
    #include <fstream>
    #include <set>
    using namespace std;
    
    const int maxLineNum = 100;
    set<string> key;             //关键字
    char ch;                     //当前读取字符
    int col, maxcol;             // col 读取当前行位置，当前行最大长度
    ifstream readfile;           //文件输入流
    char line[maxLineNum];       //读取文件一行
    char identifier[maxLineNum]; //保留读取标识符或者关键字
    void init();                 //进行初始化工作
    int getsym();                //进行词法分析
    int getch();                 //读取一个字符
    
    int main()
    { 
        cout << "请输入要分析的文件名（请放在该程序同一目录）: ";
        string filename = "";
        cin>>filename;
        readfile.open(filename);
        if (readfile.is_open() == false)
        { 
            cout << "文件不存在";
            return 0;
        }
        init();
        col = maxcol = 0;
        ch = ' ';
        while (getsym())
            ;
        return 0;
    }
    
    void init()
    { 
        key.insert("const");
        key.insert("do");
        key.insert("while");
        key.insert("if");
        key.insert("else");
        key.insert("for");
        key.insert("int");
        key.insert("char");
        key.insert("return");
        key.insert("void");
        key.insert("read");
        key.insert("write");
    }
    
    int getch()
    { 
        if (col == maxcol)
        { 
            if (readfile.peek() == EOF)
            { 
                return -1;
            }
            maxcol = col = 0;
            ch = ' ';
            while (ch != '\n')
            { 
                if (readfile.peek() == EOF)
                { 
                    line[maxcol] = 0;
                    break;
                }
                readfile.get(ch);
                line[maxcol++] = ch;
            }
        }
        ch = line[col++];
        return 0;
    }
    
    int getsym()
    { 
        int i, j, k, num;
        while (ch == ' ' || ch == 10 || ch == 13 || ch == 9)
        { 
            if (getch() == -1)
                return 0;
        }
        if (isalpha(ch))
        { 
            k = 0;
            while (isalpha(ch) || isdigit(ch))
            { 
                identifier[k] = ch;
                k++;
                if (getch() == -1)
                    return 0;
            }
            identifier[k] = 0;
            if (key.find(string(identifier)) == key.end())
            { 
                cout << "标识符 (3 , '" << identifier << "')\n";
            }
            else
            { 
                cout << "关键字 (1 , '" << identifier << "')\n";
            }
        }
        else
        { 
            if (isdigit(ch))
            { 
                k = num = 0;
                while (isdigit(ch))
                { 
                    num = num * 10 + ch - '0';
                    k++;
                    if (getch() == -1)
                        return 0;
                }
                if(isalpha(ch)||ch=='_'){ 
                    cout<<"非法标识符 (6,'"<<num<<ch<<"')\n";
                }else{ 
                    cout << "整数 (2 , '" << num << "')\n";
                }
            }
            else
            { 
                if (ch == '<')
                { 
                    if (getch() == -1)
                        return 0;
                    if (ch == '=')
                    { 
                        cout << "运算符 (4 , '<=')\n";
                        if (getch() == -1)
                            return 0;
                    }
                    else
                    { 
                        cout << "运算符 (4 , '<')\n";
                    }
                }
                else if (ch == '>')
                { 
                    if (getch() == -1)
                        return 0;
                    if (ch == '=')
                    { 
                        cout << "运算符 (4 , '>=')\n";
                        if (getch() == -1)
                            return 0;
                    }
                    else
                    { 
                        cout << "运算符 (4 , '>')\n";
                    }
                }
                else
                { 
                    if (ch == '+')
                    { 
                        cout << "运算符 (4 , '+')\n";
                    }
                    else if (ch == '-')
                    { 
                        cout << "运算符 (4 , '-')\n";
                    }
                    else if (ch == '*')
                    { 
                        cout << "运算符 (4 , '*')\n";
                    }
                    else if (ch == '/')
                    { 
                        cout << "运算符 (4 , '/')\n";
                    }
                    else if (ch == '=')
                    { 
                        cout << "运算符 (4 , '=')\n";
                    }
                    else if (ch == '(')
                    { 
                        cout << "运算符 (4 , '(')\n";
                    }
                    else if (ch == ')')
                    { 
                        cout << "运算符 (4 , ')')\n";
                    }
                    else if (ch == '.')
                    { 
                        cout << "运算符 (4 , '.')\n";
                    }
                    else if (ch == ',')
                    { 
                        cout << "分界符 (5 , ',')\n";
                    }
                    else if (ch == ';')
                    { 
                        cout << "分界符 (5 , ';')\n";
                    }
                    else if (ch == '{')
                    { 
                        cout << "分界符 (5 , '{')\n";
                    }
                    else if (ch == '}')
                    { 
                        cout << "分界符 (5 , '}')\n";
                    }
                    else
                    { 
                        cout << ch << " error!请检查代码是否拼写正确！";
                        return 0;
                    }
                    if (getch() == -1)
                        return 0;
                }
            }
        }
        return 1;
    }

### 运行截图 ###

![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxXzQxMTQ2NjUw_size_16_color_FFFFFF_t_70]

[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxXzQxMTQ2NjUw_size_16_color_FFFFFF_t_70]: https://img-blog.csdnimg.cn/20191130093457373.jpg?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3FxXzQxMTQ2NjUw,size_16,color_FFFFFF,t_70